How do I monitor brand mentions on Douyin and get new videos every day?

A brand team wants every new Douyin video that mentions its name, its flagship product or a campaign hashtag in a morning report, with the ones gaining traction on top, so they can comment, reach out to the creator or flag a problem before it spreads. Douyin's own search has a publish-time filter, but tools that use it report results from far outside the chosen window, and relevance ranking mixes old hits with new uploads. A daily search sorted by newest, cut off at the time of your last run, gives exactly that list.

The method

Use Douyin video search with sort=latest:

1. Run each keyword once a day (hourly during a launch or a crisis).
2. Keep videos whose published_at is after the last run; ignore IDs you already stored.
3. Stop paging when a page reaches older videos.
4. Rank the new ones by likes, comments and shares, and send the list on.

There is no date parameter to rely on, and none is needed: sorting by latest and cutting at a timestamp is exact.

Cost

Setup	Calls per day	Per day	Per 30 days
1 keyword, usually 1 page	1	$0.02	$0.60
10 keywords, 2 pages each	20	$0.40	$12.00
plus 2 comment pages on 3 hot videos a day	6	$0.012	$0.36

To follow how a video develops over the following days, refresh it with video detail at $0.002; that is far cheaper than searching again.

Script

import os, requests, datetime as dt

URL = "https://api.okdata.online/v1/douyin/videos/search"
HEAD = {"Authorization": f"Bearer {os.environ['OKDATA_API_KEY']}"}
since = (dt.datetime.now(dt.timezone.utc) - dt.timedelta(days=1)).strftime("%Y-%m-%dT%H:%M:%S")
params, fresh = {"query": "某品牌 新品", "sort": "latest"}, []

for _ in range(3):  # page cap
    body = requests.get(URL, headers=HEAD, params=params, timeout=60).json()
    items = body["data"]["items"]
    fresh += [v for v in items if v["published_at"] >= since]
    if not body["page"]["has_more"] or (items and items[-1]["published_at"] < since):
        break
    params["cursor"] = body["page"]["next_cursor"]
fresh.sort(key=lambda v: v["metrics"]["likes"] or 0, reverse=True)

Expect these

• Search matches loosely. Short brand names catch unrelated videos; add a product word or filter captions in code.
• Play counts are not public; rank by likes, comments and shares.
• Follower counts are not in search results. To flag big creators, look authors up with account profile.
• A page with nothing new still costs one call. Match frequency to how fast the keyword moves.

Alternatives

巨量算数 shows how search interest in a keyword moves over time, a good companion that does not list videos. Social listening suites cover Douyin with dashboards and alerts for teams that want no code. For a combined Douyin, Xiaohongshu and Weibo setup, see word-of-mouth monitoring.